NI6509 PCI I/O line impedance when powered off.

[kevdev]

We have two NI6509 PCI boards - one in each of two linux machines (CPCCs).  These two boards each have an input line attached to a single monitor point that is held to a TTL High by a pullup resistor.  So both machines are monitoring the same signal.  This is for redundancy.  This monitor circuit goes to a smoke alarm.  If an alarm occurs, a relay is closed and pulls the monitor point to ground which then causes the CPCCs to take action. 

We noticed that when one CPCC is powered off, the monitor point is no longer held to a TTL High.  It is like the NI Board input has a low impedance causing a pull down.  Measuring the line directly on the powered off board does not show a low impedance.  So I am confused.

I think I attached a schematic of the setup but don't know if I did it correctly.  If you don't see it, please use this URL: http://www.aoc.nrao.edu/~kryan/smoke-alarm-monitor.pdf

Does anyone know why the input line of a powered off board would pull the other input line that it is connected to low?

[TaranjeetHullon]

Can you elaborate how the 5V pull-up is connected ? 

[kevdev]

The 5V is supplied by one output bit of both the NI6509 boards.

[kevdev]

But the 5-VDC outputs are Wired-OR so if one output goes off, the other one is diode protected from sinking back through it.

[kevdev]

I've added a more complete schematic that also shows the output pull-up circuit.  It can also be viewed at:

http://www.aoc.nrao.edu/~kryan/smoke-alarm-monitor-circuit-complete.pdf

[TaranjeetHullon]

The schematic seems to indicate that 5V is not a real 5V supply but the output of P3.1. Is that the case ? There should be a real 5V supply exposed on the 6509 (Pin 49/99), can you use that instead ?

A typical 5V digital output behaves as a 5V supply in series with a resistance. When drawing next to 0 current, the output is close to 5V but with 24mA, the output is closer to 3.4V. 

[kevdev]

Thanks for your help by the way.  Yes,  we have 4.24 V at J4-10 and 3.4+ volts at J6-13 which is perfectly fine.  The problem is, when one of the CPCCs is powered off, that voltage (at J6-13) drops to 1.3 V making the other CPCC think that there is a smoke alarm.  Yet when I measure resistance at J6-13, I see 5.24 K ohms which is what one would expect from isolation and pull-down resistors.  This indicates to me that the 6509 input line is of high impedance.  So what could be pulling that voltage down?  With one CPCC off, we still see 4.21 V at J4-10 so I would expect about 3.1 V at J6-13 not 1.3.

[TaranjeetHullon]

That is intriguing.

I believe (NI folks needs to confirm), based on the spec sheet which indicates an input voltage of 0V to Vcc, that the internals of the DIO input pins have a typical ESD protection configuration, which is a diode to the supply voltage and another to ground. If I'm right, this means when powered off, any voltage above 0.5V on the pins will cause the diode to conduct. 

Any chance you can throw in a simple buffer in the circuit ?

[kevdev]

We had just started speculating about some sort of diode issue that would cause conduction with a voltage applied and your idea supports that.  It would be nice if NI gave us a schematic of those lines.

Our plan is to remove the two pull-down resistors altogether and change R48 and R90 to 10K.  This is all surface mount stuff on a printed circuit board so we can't easily add anything.  I'll let you know how it turns out.

Thanks for you thoughts, they're much appreciated.